Skip to contents

\(IE = ab\) with \(s_{ab} = \sqrt{b^2 s_a^2 + a^2 s_b^2}\). Reported because it is what this quantity is classically paired with, but the product of two normals is not normal and is skewed when both coefficients are near zero, so the test is conservative exactly where mediation is most in doubt; a bootstrap is the better instrument there. Without both standard errors the test is skipped rather than faked.

Usage

morie_indirect_effect_sobel(a, b, se_a = NULL, se_b = NULL, alpha = 0.05)

Arguments

a, b

Path coefficients.

se_a, se_b

Their standard errors; both needed for a test.

alpha

Significance level.

Value

Named list with estimate, se, statistic, p_value, ci_low, ci_high.

Details

Mirrors morie.fn.abind.

References

Sobel ME (1982). Asymptotic confidence intervals for indirect effects in structural equation models. Sociological Methodology, 13, 290-312.

Examples

morie_indirect_effect_sobel(0.5, 0.4, 0.1, 0.08)$p_value
#> [1] 0.000406952